ieee transactions on power electronics, vol. 27, no. …ho et al.: practical design and...

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012 2835

Practical Design and Implementation Procedure of anInterleaved Boost Converter Using SiC Diodes for

PV ApplicationsCarl Ngai-Man Ho, Member, IEEE, Hannes Breuninger, Member, IEEE, Sami Pettersson, Member, IEEE,

Gerardo Escobar, Senior Member, IEEE, Leonardo Augusto Serpa, Member, IEEE,and Antonio Coccia, Member, IEEE

Abstract—The implementation of an interleaved boost converter(IBC) using SiC diodes for photovoltaic (PV) applications is pre-sented in this paper. The converter consists of two switching cellssharing the PV panel output current. Their switching patterns aresynchronized with 180◦ phase shift. Each switching cell has a SiCSchottky diode and a CoolMOS switching device. The SiC diodesprovide zero reverse-recovery current ideally, which reduces thecommutation losses of the switches. Such an advantage from theSiC diodes enables higher efficiency and higher power density ofthe converter system by reducing the requirement of the coolingsystem. This paper presents also an optimization study of the sizeand efficiency of the IBC. Based on 1) the steady-state character-istic of the topology; 2) the static and dynamic characteristics ofthe switching cells; 3) the loss model of the magnetic components;and 4) the cooling system design, the paper provides a set of designcriteria, procedures, and experimental results for a 2.5 kW IBCprototype using SiC diodes.

Index Terms—Diode, interleaved boost converter (IBC), MOS-FET, power semiconductor, photovoltaic (PV), silicon carbide(SiC).

I. INTRODUCTION

S ILICON carbide (SiC) represents a breakthrough in silicontechnology because it allows a larger energy gap. SiC isclassified as a wide-band-gap (WBG) material, and it is becom-ing the mainstream material for power semiconductors [1], [2].Among the different types of power semiconductors, the powerdiode was the first device to adopt SiC technology, whichwas commercialized years ago. The main advantages are thehigh-breakdown voltage and the small reverse-recovery current.Some research has proven that SiC Schottky diodes are superiorto Si-based diodes in device characteristics [3]–[5]. As a result,higher efficiency and higher power density can be brought topower electronic systems in different applications [6]–[8].

Manuscript received August 9, 2011; revised October 8, 2011; acceptedNovember 15, 2011. Date of current version March 16, 2012. Part of the workdescribed in this paper has been published in ECCE-Asia 2011. Recommendedfor publication by Associate Editor T. Shimizu.

The authors are with ABB Switzerland, Ltd., Corporate Research, Baden-Dättwil 5405, Switzerland (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TPEL.2011.2178269

Fig. 1. Dual-stage topology for PV inverter using IBC.

The market for residential photovoltaic (PV) inverters is be-coming highly competitive. PV manufacturers are competingto increase the efficiency for every 0.1%. From the maximumpower point tracking (MPPT) algorithm point of view, the exist-ing methods, such as perturbation and observation (P&O) andincremental conductance (IncCond), can track the maximumpower point properly and the dynamic response is good enoughto deal with changes in temperature and irradiation [9], [10].From the hardware point of view, there are 2 DoFs in an inverterdesign used to improve the efficiency, namely semiconductorand topology. As the topology is limited by the issue of com-mon mode voltage, the options of transformerless topologiesare limited [11]. With regard to the semiconductor, high voltageand low current ratings for residential PV inverters are required.The commercialized SiC diodes are acceptable in this particu-lar application from the electrical performance point of view.However, it is well known that SiC increases the overall cost ofcomponents. Moreover, a single diode replacement without anyoptimization cannot effectively improve the system efficiency.Instead it may prolong the payback time from electricity savingsto compensate for the cost of the PV inverter. Thus, the rightselection of topology and peripheral devices, such as switches,passive devices, and cooling systems, is important in order tomaximize the benefits of using SiC diodes in a power electronicssystem.

Fig. 1 shows a typical dual power stage single-phase trans-formerless PV inverter [11]–[14]. It is formed by two convert-ers: 1) a dc–dc converter for MPP tracking and dc-link volt-age stabilizing (referred to as a preregulator) and 2) a dc–ac inverter for injecting ac power into the grid. It is a wellknown fact that the I–V characteristics of PV panel vary withthe environment, such as temperature and irradiation. For gen-eral purpose single-phase inverter for low-power residential PV

0885-8993/$26.00 © 2011 IEEE

2836 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 6, JUNE 2012

TABLE ISPECIFICATIONS OF A BOOST CONVERTER FOR SINGLE-PHASE PV

INVERTERS PREREGULATION

installation, the variation range of panel MPP voltage is typi-cally from 125 to 650 V. And the absolute open-circuit voltagecan be as high as 800 V. Although the inverter does not convertenergy at this voltage, the breakdown voltage of semiconduc-tors and capacitors has to deal with the voltage level. Besides,the dc-link voltage Vdc has to be higher than the peak of thegrid voltage, e.g., 400 V. Thus, there are two operating modesin the energy converting process to reduce the losses in the PVinverter. They are as follows:

Mode 1: The PV panel output voltage is from 125 to 400 V,the dc–dc converter boosts up the dc-link voltage to 400 V andit handles the MPPT function.

Mode 2: The PV panel output voltage is from 400 to 650 V,the dc–dc converter bypasses by the diode, DB. The PV panelis directly connected to the inverter and the inverter takes overthe MPPT function.

Moreover, both power stages have to operate at/or higherthan 16-kHz switching frequency for residential PV inverters,in order to avoid acoustic noise generated by the inductors. An-other limitation of the converter design is that the input currentvariation cannot be too large, otherwise the operating point isnot stable at the MPP of PV. Table I summarizes the typicaloperating conditions of residential PV inverters.

This paper presents a practical design and implementationprocedure for an interleaved boost converter (IBC) using SiCSchottky diodes in a residential PV preregulator application. Itmust be noted that this represents an example of the use of themethod and procedure. It can be extended to optimize the dc–acinverter. The design goal is to maximize the efficiency in thesystem and the design criteria are in agreement with the typicalspecification of single-phase PV inverters in Table I. The de-sign procedure is based on the basic analysis of the steady-statecharacteristics of the topology and the semiconductor switchingbehavior. The further optimization for the passive devices andcooling system can be obtained based on the previously analyzedresults. Fig. 2 shows the flow chart of the design steps. Experi-mental results in a 2.5 kW IBC prototype using SiC diodes areprovided to show the performance of the optimized prototype.

II. REVIEW OF STEADY-STATE CHARACTERISTICS OF IBC

IBC consists of “n” single boost converters connected in par-allel, i.e., n-phase, to share the handled power. In this paper,the two-phase IBC, shown in Fig. 1, is considered. Notice thatthe results can be easily extended to n-phases. In the case of

Fig. 2. Design procedure for prototyping the optimized IBC.

the two-phase IBC, the two switches S1 and S2 are controlledwith 180◦ phase shift. By using this circuit structure and mod-ulation scheme, the advantage of antiphase ripple cancellationof both inductors can be achieved. The amplitude of the inputcurrent ripple is smaller compared to a single boost converter.This makes the topology very attractive for PV preregulator ap-plications. However, as the voltage of the PV string can reach800 V, then in a purely silicon (Si) design, the converter suf-fers from reverse-recovery losses and commutation losses. SiCdiodes are used to solve this problem because they have prac-tically negligible reverse recovery and have a high-breakdownvoltage.

In principle, the IBC is formed by two independent boostswitching units. For each boost switch unit, there are two switch-ing states.

1) Switch is ON. The current in the inductor starts to rise,while the diode is blocking.

2) Switch is OFF. The inductor starts to discharge and transferthe current via the diode to the load.

The following are the most important parameters used tostudy the steady-state characteristic of the IBC.

1) Boost Ratio: The boosting ratio of the converter is a func-tion of the duty ratio. It is the same as in a conventional boostconverters. It is defined by the duty ratio

VdcVin

=1

1 − D (1)

where Vdc is the output voltage, Vin is the input voltage, and Dis the duty ratio.

2) Input Current: The input current can be calculated by theinput power and the input voltage

Iin =PinVin

. (2)

3) Inductor Current Ripple Peak-to-Peak Amplitude: Theinductor current ripple peak-to-peak amplitude can be

HO et al.: PRACTICAL DESIGN AND IMPLEMENTATION PROCEDURE OF AN INTERLEAVED BOOST CONVERTER 2837

determined by

ΔIL1,L2 =Vin · Dfsw · L

(3)

where fsw is the switching frequency and L is the inductance.4) Relationship Between Input Current Ripple Peak-to-Peak

Amplitude and Inductor Current Ripple Peak-to-Peak Ampli-tude: As with most interleaved converters, the minimum inputripple occurs at a duty cycle of 0.5. This is due to the 180◦ phasedifference between the two boost cells [15], [16]. There are twooperating modes which can be defined by the inductor currentbehaviors.

1) Mode 1 (D > 0.5): Over a specific period of time thecurrent in both inductors rises.

2) Mode 2 (D < 0.5): Over a specific period of time bothinductors discharge.

Consequently, the input current ripple peak-to-peak ampli-tude is given by

ΔIin =Vdc − 2 · Vin

fsw · L·{

−D, if D ≤ 0.51 − D, if D > 0.5. (4)

5) Operating Currents in Semiconductors: Fig. 3 shows thetypical waveforms of one leg of the IBC and the semiconductorloss distribution in one switching cycle. Ideally, the input currentis evenly shared by the two switching cells. It means that halfof the input current is flowing in each leg, i.e., Iin /2. Thus

iD,off = iF,on =Iin + ΔIL1,L2

2(5)

iD,on = iF,off =Iin − ΔIL1,L2

2(6)

where iD,off and iD,on are drain currents of the MOSFETs atturn-off and turn-on transients, respectively. iF,off and iF,on areoperating currents of the diodes at turn-off and turn-on tran-sients, respectively.

The RMS values of the current flowing through the semicon-ductors are given by

ID,RMS =

√√√√(

i2D,on +iD,on · ΔIL1,L2

2+

ΔI2L1,L23

)· D (7)

IF,RMS =√√√√(i2F,off +

iF,off · ΔIL1,L22

+ΔI2L1,L2

3

)· (1−D). (8)

Based on (1)–(8), the inductance, the inductor losses, and thesemiconductors losses can be determined.

III. INDUCTORS

In modern power electronics systems, magnetic componentsplay a very important role with regard to storage and filtering. In

Fig. 3. Typical waveform of one leg of the IBC at rated power.

the operating principle of a boost converter, the inductor is usedto transform the energy from the input voltage to the inductorcurrent and to convert it back from the inductor current to theoutput voltage. Fig. 1 shows that there are two independentinductors in an IBC. In principle, the two inductors are identicalin order to balance the current in the two boost cells. In thissection, the design and implementation of the inductors arepresented as well as the loss model of the inductors providedfor the overall system loss estimation.

A. Determination of Inductor Value

According to the specifications in Table I, the boost converteroperates when the input voltage is in the range from 125 to400 V. Based on (1), the corresponding duty ratio range is 0 to0.69. Fig. 4 shows a typical profile of the input ripple peak-to-peak amplitude against the duty ratio in an IBC, which can bedetermined using (4). It shows that the maximum amplitude ofthe input current ripple occurs at 0.25 duty ratio in the specifiedoperating range. Thus, based on Table I and from (1) to (4), theinductor value can be determined by

L =

[Vin · D ·

(2 − (1 − D)−1

)fsw · 0.1 · (Pin, max/Vin,min)

]D=0.25

. (9)

Finally, the resulting inductance per inductor is 1.5 mH andtwo inductors are used in the IBC.


Fig. 4. Input current ripple current versus duty cycle.

TABLE IISUMMARY OF INDUCTOR DESIGN

B. Implementation

The main assembling components of an inductor are coreand wire. The material of the core for the operating frequencycan be ferrite, iron powder, and Metglas. Among the materials,Metglas provides a very high saturation flux density due to theiron-based amorphous alloy 2605SA1 [17]. Thus, the size andcore loss can be minimized compared to the other materials. Inthe paper, the Metglas C-Core AMCC-20 and Litz wire (25 mm× 0.355 mm) have been selected based on the loss and sizeconsideration. Two inductors have been implemented based onthe design procedures in [18]. The practical design considerationwill be discussed in Section VII. The summary of the inductordesign and implementation is shown in Table II.

C. Inductor Loss Model

The losses in practical inductors can be divided in two maincategories: copper loss and core loss. The copper loss dependson the length, thickness, and material of the wire. The coreloss mainly appears in the air gap of the magnetic components.The loss model and equations are based on [18]. Please refer to

TABLE IIIPARAMETERS OF INDUCTOR LOSSES

Table III where all symbols used in the following equations aredefined.

1) Copper Loss: The resistance per unit length Runit, of con-ductor is given by

Runit =ρ20 [1 + α20(Tmax − 20)] · n

b · c · fw. (10)

Notice that a, b, c, and d are the geometrical values for thecore. “a” is the width of the core leg, “b” and “c” are the widthand height of the window, respectively, the “c” is the thickness ofthe core. The detail geometry is given in [18]. Besides, the max-imum operation temperature Tmax , has been taken into accountin (10). Consequently, the result for the copper loss considersthe temperature at 80 ◦C in the whole operating range. This isthe worst case of the inductor temperature.

For the calculation of the total resistance, the mean turnlength, MTL, is approximated as follows:

MTL = 2 · (a + 2b + d). (11)

Therefore, the total resistance Ωtot , is given by

Ωtot = Runit · MTL · n. (12)Furthermore, a Litz wire was applied to decrease skin and

proximity effects. The copper loss can be calculated with thecurrent per leg as follows:

Pcu = I21,2 · Ωtot . (13)2) Core Loss: The ac flux in the air gap Bac , is given by the

following equation:

Bac =0.4 · π · n · ΔIL

2 · lg. (14)

Consequently, the core loss Pcore , can be calculated using the“Steinmetz” equation, which is provided by the manufacturer’sapplication note [18], and is given by

Pcore = 6.5 · f 1.51 · B1.74ac · wt. (15)According to the data sheet, the temperature condition for

(15) is 25 ◦C. It also shows that the core loss variation is lower


Fig. 5. Inductor losses at 80 ◦C.

than 5% when the temperature changes from 25 to 100 ◦C. Thus,the assumption that power loss is independent of temperaturecan be made.

With this result of copper loss, the total loss of two inductorsPtot , can be estimated as

Ptot = 2 · (Pcu + Pcore). (16)

The results of the losses are shown in Fig. 5, which are basedon the parameters in Table III. The copper loss, core loss, andtotal loss of the inductors at the thermal design point are 3,8.7, and 23.5 W, respectively. The graph also shows that thecore loss is the dominant part of the total losses and could bereduced by using less turns, smaller current ripple, or longerair gap. However, this has not been done in the present design.This is because the specification of the input ripple leads to afixed inductance value, which causes fixed current ripple perleg. Also the number of turns cannot be decreased because ofthe saturation issue of the core. A tradeoff thus arises. If lowerlosses are to be achieved, then a bigger core has to be used toreduce the number of turns, but consequently the size of the corewill increase as well.

IV. SEMICONDUCTORS EVALUATION

Semiconductor characterization is the first step in prototypinga power electronic system. Based on the data extracted from thesemiconductors, the conduction losses and switching losses of aswitching cell operating in a system can be estimated [19], [20].Moreover, it is important to optimize the system by selectingthe most suitable semiconductor devices and gate drive circuits.In this paper, CoolMOS and SiC Schottky diodes were selectedas the active switch and diode, respectively. Table IV shows thepart numbers and parameters of these devices. The advantagesof the CoolMOS are low on-state resistance at low-junctiontemperature and no “tail” current during the turn-off transients.These result in relatively lower conduction loss and turn-offswitching loss compared to IGBTs. Moreover, the SiC diodesbring low reverse-recovery losses and low commutation losses inthe switching cell [21], [22], which is advantageous for systemswith high-efficiency requirements.

TABLE IVPARAMETERS OF THE EVALUATED DEVICES

This section demonstrates the experimental results of the de-vice characterization. It includes the output characteristics ofthe CoolMOS, the forward characteristic of the SiC diode, andthe switching loss chart of one switching cell. An estimated lossbreakdown for the semiconductors is included as well.

A. Static Characteristics of Semiconductors

The aim of the static characteristics measurements is to de-termine the conduction loss of the semiconductors, which willbe used in the power electronics system. The Tektronix 371 ACurve Tracer was used to extract the parameters from the semi-conductor devices.

Fig. 6 shows the output characteristics of the CoolMOS at dif-ferent junction temperatures. The on-state resistances are 108,156, and 250 mΩ at 25, 75, and 125 ◦C, respectively. This impliesthat the conduction loss of the CoolMOS dramatically increasesat high-junction temperature. Fig. 7 shows the forward charac-teristics of the SiC Schottky diode as well as the CoolMOS, thediode forward characteristics are temperature dependent withpositive coefficient. According to the static characteristics ofthe devices, 75 ◦C is a suitable junction temperature for bothdevices due to the low conduction loss in the semiconductors.In addition, the results show that the static characteristics ofthe components are close to the given results in the device datasheets.

B. Dynamic Characteristics of Semiconductors

Energy loss information in a switching cell can be extractedby double pulse test setup [23]. It simulates the switching actionsof the switching cells in a power electronics system. Generally,the turn-on and turn-off switching loss of the switch and theturn-off switching loss of the diode are considered in the lossmeasurements as they dominate the loss during transients inthe devices. Generally, the inductive switching loss informationis not given by the semiconductor data sheet. This is becausedifferent combinations of active device and diode will result indifferent dynamic behaviors. It is hard to provide informationby calculation, to estimate the switching losses in the semicon-ductors. Double pulse test is the usual way to obtain the accurateswitching loss information before designing a converter.

Fig. 8 shows the experimental switching loss chart forthe switching cell which is listed in Table IV. The testing


Fig. 6. CoolMOS output characteristic with different junction temperatures.

Fig. 7. SiC diode forward characteristic with different junction temperatures.

environment is simulated with an IBC operating, at 400 Vdc-link voltage and 75 ◦C juncture temperature. The switch-ing action time intervals are illustrated in Fig. 3. The chartshows that the reverse-recovery loss in the diode is very low,around 40 μJ at 10 A operation. The reason for this is theuse of the SiC Schotkky barrier diodes that brings down thereverse recovery. Meanwhile, it reduces the turn-on transitiontime of the CoolMOS. Thus, the turn-on switching loss is lowas well. Besides, the short turn-off transient time is a signifi-cant advantage of CoolMOS. It is proven in the measurementsthat, the turn-off loss of the CoolMOS is very low compared toIGBTs [21], [22].

Fig. 8. Switching loss chart.

C. Semiconductor Loss Breakdown at Thermal Design Point

As previously mentioned, the main purpose of having thestatic and dynamic characteristics of the semiconductors is touse this information to predict the semiconductor loss in a powerelectronics system. Then, an appropriate cooling system can bedesigned based on the losses breakdown at the thermal designpoint, i.e., Pin = 2.5 kW, Vin = 125 V, and D = 0.6875. Noticethat the semiconductors produce the highest loss at this point.

1) Conduction Losses: L, ΔIL1,L2 , and iD,on are determinedin (9), (3), and (6), respectively. Out of these, ID,RMS is obtainedby (7). The conduction loss of the CoolMOS is determined by

PM con = ID,RMS · VDS (17)

where VDS can be determined using Fig. 6 for the correspondingID,RMS .

Similarly, IF,RMS is obtained by (8). The conduction loss ofthe diode is determined by

PD con = IF,RMS · VF . (18)

where VF can be determined using Fig. 7 for the correspondingID,RMS .

2) Switching Losses: The switching energy losses in the de-vices are simply determined using Fig. 8 for the correspondingcurrents. The equations are

PM on = EM,on(iD,on) · fsw (19)PM off = EM,off (iD,off ) · fsw (20)PD rr = ED,rr(iF,off ) · fsw . (21)

Table V shows the estimated loss breakdown of the switchingcell operating in the IBC at 16-kHz switching frequency. It isobserved that the reverse-recovery loss (PD rr) in the SiC diodeis not significant, and the switching losses in the CoolMOS arequite low.


TABLE VLOSS BREAKDOWN OF SEMICONDUCTORS IN IBC

Fig. 9. Simplified thermal resistance network.

Fig. 10. Thermal simulation of designed cooling system.

V. COOLING SYSTEM

There are generally two main components in a cooling sys-tem for air-force convection, namely a heat sink and a fan.An aluminum 42-fin heat sink and two cooling fans, SUNONKDE1204PKV2, are used in the system. The thermal model ofthe cooling system has been created using “Qfin4” software.Based on the semiconductors J-C thermal resistances shown inTable IV and the loss of each device in Table V, the heat sinklength can be determined. Fig. 9 shows the simplified thermalresistance network. The graphical simulated results are shown inFig. 10. The figure shows the placement of the semiconductors,the dimensions of the heat sink and cooling fans, and the heat dis-tribution. According to the simulated results, the maximum junc-tion temperature of the CoolMOS in the system operating in thecritical conditions is 76.2 ◦C. Notice that it is approximately thesame as the designed junction temperature of the CoolMOS inSection IV.

Fig. 11. Photograph of the optimized IBC prototype.

TABLE VITEMPERATURES IN IBC

VI. EXPERIMENTAL VERIFICATIONS

A laboratory prototype has been built based on the designprocedures presented in the paper. The prototype is shown inFig. 11. It shows all boards and components in the converter.The dimensions of the prototype are L150 × W170 × H150mm.

The loss prediction in Sections III and IV for the inductors andthe semiconductors in the critical conditions, 125-V input volt-age and 2.5-kW output power, are 23.5 and 44.7 W, respectively.As a result, the measured system loss was 69 W. It is 0.8 W higherthan the prediction, which is not significant. The reasons for thisdifference are as follows. First, the ambient temperature was notat 50 ◦C, thus the semiconductors were not working at the de-sired junction temperature of 75 ◦C. As a result, the loss of MOS-FET is slightly different from the calculated values. Second, thelosses of sensors, relays, and other passive devices have not beentaken into account in the simulation. Nevertheless, the measuredtemperatures are shown in Table VI. The temperature informa-tion shows that the cooling system is designed properly. Themeasured case temperature of the CoolMOS has around 22 ◦Cof difference with respect to the simulated result in Fig. 9, whilethe ambient temperatures have 23 ◦C of difference.

Fig. 12 shows the switching waveforms when the systemis working in the critical conditions. It shows that the currentwaveforms of iL1 and iL2 are 180◦ out of phase, and they arealmost in the even sharing mode. In fact, the distribution ofthe currents in the two switching cells is highly dependent onthe parameters of the inductors and semiconductors. Besides,the input current is the sum of the two inductor currents. Thecurrent ripple amplitude is around 3.2 A, which matches (3).Fig. 13 shows the waveforms when the system is operating


Fig. 12. Measured current waveforms of IBC at 2.5 kW, 125 V and D = 0.6875.

Fig. 13. Measured current waveforms of IBC at 2.5 kW, 200 V and D = 0.5.

at full load and the duty ratio is 0.5. The input current rippleamplitude is zero because of the ripple cancellation between thetwo inductor currents.

In order to verify the simulated result in the optimizing pro-cess, Fig. 14 shows the simulated and measured efficiencies inthe completed PV operating range. In the simulation, the semi-conductor losses are based on the experimental results shownin Figs. 6–8. The loss estimating program simulates the IBCoperating at different input voltages and output powers. Thecharts show that the results are very similar in most regions,only the efficiency at low power and high-input voltage regionare higher than 0.5% difference. The reasons for the differenceare that when the output power is very low, the measuring errorbecomes significant in the efficiency calculation. Moreover, theinput parameters of the simulation are in constant temperaturecondition, however, in practice, the temperature is changing si-multaneously with the output power. Although the simulationresults are slightly different with respect to the experimentalresults, they provide a very good overview for the system be-haviors before the experiment is carried out.

Fig. 15 shows the European efficiency of the IBC operatingin a PV system. The solid line is the optimized IBC using Cool-MOS devices and SiC diodes. In contrast, another optimizedIBC using commercial IGBTs and SiC diodes has been imple-mented and measured. The dash line shows the result of thislatter system. It can be seen that the CoolMOS device provides1% efficiency improvement compared to the IGBT, when SiCdiodes are applied to the IBC. This illustrates the importance ofan appropriate semiconductor selection.

VII. DISCUSSIONS

A. Optimal Inductor Design

Loss, size, and cost of an inductor are always a tradeoff inthe inductor design. There are no restricting criteria to guideengineers to design inductors to achieve an optimum design.The design should depend on the available physical dimensionsand the loss optimization. Typically, there are two degrees offreedom in an inductor design to fit into the system, namely


Fig. 14. Efficiencies in the whole PV system operating range. (a) Simulated.(b) Measured.

Fig. 15. European efficiency of the IBC.

Fig. 16. Inductor design curves.

turns optimization and core selection. Fig. 16 shows the lossresponse of different inductor designs which can be used in thePV preregulator in the paper. All design points in the chart werecalculated from (10) to (16). The optimal number of turns of theAMCC16B, AMCC20, and AMCC25 cores is 68, 63, and 70turns, respectively. Based on the analysis, AMCC25 is the mostsuitable core due to the lowest loss. However, actual physicaldimensions need to be considered in the practical design. As theinductors were to be located at the air flow output of the heatsink, which is shown in Fig. 11, then the total length of twocores could not be longer than the length of the heat sink (155mm). Unfortunately, the total length of two AMCC25 is longerthan the heat sink length. Therefore, AMCC20 core has beenselected in this design, since the length of each core is 72 mmaccording to the data sheet [18]. With the local optimization,AMCC20 core with 63 turns is the theoretical design point ofthe inductor. However, the air gap should be minimized in theinductor, since a large air gap generates extra power losses and arise in the conductor temperature, which is caused by the air gapfringing flux [24]. Therefore, a minimum numbers of turns andthe shortest air gap were designed in the inductor. Finally, theloss difference between practical and theoretical inductor designpoint was 2 W based on (16). This difference is, however, notsignificant from the full system efficiency point of view.

B. Optimal Switching Frequency Selection

It is well known that the switching losses in the systemcan be reduced by using SiC semiconductors. Therefore, highefficiency can be achieved with high-switching frequency inconverters. The switching frequency optimization of the IBCis shown in Fig. 17. It illustrates that the optimal switchingfrequency is 11 kHz. However, it has been mentioned thatthe minimum switching frequency of the application must be16 kHz to avoid acoustic noise. Thus, the optimal switchingfrequency was fixed at 16 kHz due to the lowest total loss in thesystem within the usable frequency bandwidth.


Fig. 17. Losses in IBC with switching frequency variation.

VIII. CONCLUSION

This paper has presented a complete design and implemen-tation procedure for the IBC prototype using SiC diodes. Thedesign criteria were based on four aspects: topology, semicon-ductors, magnetic devices, and cooling system. The steady-statecharacteristics of the IBC have been studied and the semicon-ductor losses have been experimentally obtained. Based on this,the optimized cooling system was designed to dissipate effec-tively the semiconductor losses. Moreover, the loss model of themagnetic devices was determined, thus the overall system’s sizeand efficiency could be further optimized. The experimental re-sults were similar to the simulated results in terms of junctiontemperature and efficiency. In conclusion, the converter withCoolMOS devices and SiC diodes is very suitable for PV pre-regulator applications because of the minimized system loss andsize reductions.

REFERENCES

[1] M. Bhatnagar and B. J. Baliga, “Comparison of 6 H-SiC, 3 C-SiC, andSi for power devices,” IEEE Trans. Electron Devices, vol. 40, no. 3,pp. 645–655, Mar. 1993.

[2] Q. Zhang, R. Callanan, M. K. Das, S.-H. Ryu, A. K. Agarwal, and J.W. Palmour, “SiC power devices for microgrids,” IEEE Trans. PowerElectron., vol. 25, no. 12, pp. 2889–2896, Dec. 2010.

[3] A. Elasser, M. H. Kheraluwala, M. Ghezzo, R. L. Steigerwald, N. A. Evers,J. Kretchmer, and T. P. Chow, “A comparative evaluation of new siliconcarbide diodes and state-of-the-art silicon diodes for power electronicapplications,” IEEE Trans. Ind. Appl., vol. 39, no. 4, pp. 915–921, Jul.2003.

[4] M. M. Hernando, A. Fernández, J. Garcı́a, D. G. Lamar, and M. Rascón,“Comparing Si and SiC diode performance in commercial AC-to-DC rec-tifiers with power-factor correction,” IEEE Trans. Ind. Electron., vol. 53,no. 2, pp. 705–707, Apr. 2006.

[5] B. Ozpineci and L. M. Tolbert, “Characterization of SiC schottky diodesat different temperatures,” IEEE Power Electron. Lett., vol. 1, no. 2,pp. 54–57, Jun. 2003.

[6] G. Spiazzi, S. Buso, M. Citron, M. Corradin, and R. Pierobon, “Perfor-mance evaluation of a Schottky SiC power diode in a boost PFC applica-tion,” IEEE Trans. Power Electron., vol. 18, no. 6, pp. 1249–1253, Nov.2003.

[7] A. M. Abou-Alfotouh, A. V. Radun, H. Chang, and C. Winter-halter, “A 1-MHz hard-switched silicon carbide DC–DC converter,”IEEE Trans. Power Electron., vol. 21, no. 4, pp. 880–889, Jun.2006.

[8] B. Ozpineci, M. Chinthavali, A. Kashyap, L. M. Tolbert, and A. Mantooth,“A 55 kW three-phase inverter with Si IGBTs and SiC Schottky diodes,”IEEE Trans. Ind. Appl., vol. 45, no. 1, pp. 278–285, Jan./Feb. 2009.

[9] Y.-H. Ji, D.-Y. Jung, J.-G. Kim, J.-H. Kim, T.-W. Lee, and C.-Y. Won, “Areal maximum power point tracking method for mismatching compensa-tion in PV array under partially shaded conditions,” IEEE Trans. PowerElectron., vol. 26, no. 4, pp. 1001–1009, Apr. 2011.

[10] A. K. Abdelsalam, A. M. Massoud, S. Ahmed, and P. N. Enjeti,“High-performance adaptive perturb and observe MPPT technique forphotovoltaic-based microgrids,” IEEE Trans. Power Electron., vol. 26,no. 4, pp. 1010–1021, Apr. 2011.

[11] Q. Li and P. Wolfs, “A review of the single phase photovoltaic module in-tegrated converter topologies with three different DC link configurations,”IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1320–1333, May 2008.

[12] Y. Fang and X. Ma, “A novel PV microinverter with coupled inductors anddouble-boost topology,” IEEE Trans. Power Electron., vol. 25, no. 12,pp. 3139–3147, Dec. 2010.

[13] D.-Y. Jung, Y.-H. Ji, S.-H. Park, Y.-C. Jung, and C.-Y. Won, “Interleavedsoft-switching boost converter for photovoltaic power-generation system,”IEEE Trans. Power Electron., vol. 26, no. 4, pp. 1137–1145, Apr. 2011.

[14] B. Yang, W. Li, Y. Zhao, and X. He, “Design and analysis of a grid-connected photovoltaic power system,” IEEE Trans. Power Electron.,vol. 25, no. 4, pp. 992–1000, Apr. 2010.

[15] Y.-C. Hsieh, T.-C. Hsueh, and H.-C. Yen, “An interleaved boost converterwith zero-voltage transition,” IEEE Trans. Power Electron., vol. 24, no. 4,pp. 973–978, Apr. 2009.

[16] L. Huber, B. T. Irving, and M. M. Jovanovic, “Open-loop control methodsfor interleaved DCM/CCM boundary boost PFC converters,” IEEE Trans.Power Electron., vol. 23, no. 4, pp. 1649–1657, Jul. 2008.

[17] N. DeCristofaro, “Amorphous metals in electric-power distribution appli-cations,” MRS Bull., vol. 23, no. 5, pp. 50–56, May 1998.

[18] “Power factor correction inductor design for switched mode power sup-plies using Metglas Powerlite C-Cores,” Metglas Application Guides.Available: http://www.metglas.com/downloads/apps/pfc.pdf.

[19] Y. Ren, M. Xu, J. Zhou, and F. Lee, “Analytical loss model of powerMOSFET,” IEEE Trans. Power Electron., vol. 21, no. 2, pp. 310–318,Mar. 2006.

[20] Y. Xiong, S. Sun, H. Jia, P. Shea, and Z. J. Shen, “New physical insights onpower MOSFET switching losses,” IEEE Trans. Power Electron., vol. 24,no. 2, pp. 525–531, Feb. 2009.

[21] Z. Liang, B. Lu, J. D. van Wyk, and F. C. Lee, “Integrated CoolMOSFET/SiC-diode module for high performance power switching,” IEEETrans. Power Electron., vol. 20, no. 3, pp. 679–686, May 2005.

[22] L. Lorenz, G. Deboy, and I. Zverev, “Matched pair of CoolMOS transistorwith SiC-Schottky diode–advantages in application,” IEEE Trans. Ind.Appl., vol. 40, no. 5, pp. 1265–1272, Sep./Oct. 2004.

[23] C. Ho, F. Canales, A. Coccia, and M. Laitinen, “A circuit-level analyticalstudy on switching behaviors of SiC diode at basic cell for power convert-ers,” in Proc. IEEE Ann. Meeting Ind. Appl. Soc., Oct. 2008, pp. 1–8.

[24] J. Fletcher, B. Williams, and M. Mahmoud, “Airgap fringing flux reductionin inductors using open-circuit copper screens,” IEE Proc.-Electr. PowerAppl., vol. 152, no. 4, p. 990-996, Jul. 2005.

Carl Ngai-Man Ho (M’07) received the B.Eng. andM.Eng. double degrees and the Ph.D. degree in elec-tronic engineering from the City University of HongKong, Kowloon, Hong Kong, in 2002 and 2007, re-spectively. His Ph.D. research was focused on the de-velopment of dynamic voltage regulation and restora-tion technology.

From 2002 to 2003, he was a Research Assis-tant at the City University of Hong Kong. From 2003to 2005, he was an Engineer at e.Energy Technol-ogy Ltd., Hong Kong. In May 2007, he joined ABB

Corporate Research, Ltd., Baden-Dättwil, Switzerland, where he is currently aPrincipal Scientist and Project Manager of PV inverter research projects. Heowns several patents in the area of energy saving and renewable energy conver-sion. His research interests include renewable energy conversion technologies,power quality, modeling and control of power converters, and characterizationof wide bandgap power semiconductor devices and their applications.


Hannes Breuninger (M’09) received the Diplomadegree in mechatronical engineering from theKarlsruher Institute of Technology (KIT), Karlsruhe,Germany, in 2011. For his diploma thesis he focusedon control systems for automotive clutches to reducejudder.

From 2009 to 2010, he was an Intern at ABBCorporate Research, Ltd., Baden-Dättwil, Switzer-land. His tasks implied design, implementation andperformance verification of dc–dc converters for PVapplications. Since 2011, he has been a Testing Engi-

neer with an automobile supplier, Brose, Hallstadt, Germany, and is responsiblefor test coordination and validation processes in the development department.His studies were focused on power electronic systems and feedback control sys-tems. He worked on design and implementation topics for multilevel convertersin his student research project.

Sami Pettersson (M’05) received the M.Sc. andD.Sc. degrees in electrical engineering from TampereUniversity of Technology, Tampere, Finland, in 2004and 2009, respectively. His research topic at theuniversity was shunt four-wire active power filtertopologies and their digital control methods.

From 2003 to 2004, he was a Research Assistant,and from 2004 until 2008, a Research Engineer atthe Institute of Power Electronics, Tampere Univer-sity of Technology. In December 2008, he joined thePower Electronics Systems Group at ABB Corporate

Research, Baden-Dättwil, Switzerland, where he is currently a Principal Sci-entist and Project Manager of a research project related to renewable energyconversion and industrial motor drive systems. His research interests includemodeling, design and optimization of power converter topologies and systems,power quality, as well as control of power converters.

Gerardo Escobar (SM’08) received the B.Sc. de-gree in electromechanical engineering (speciality inelectronics) and the M.Sc. degree in electrical engi-neering (speciality in automatic control), both fromthe Engineering Faculty of the National University ofMexico, Mexico, in 1991 and 1995, respectively, andthe PhD degree from the Signals and Systems Lab.LSS-SUPELEC, Paris, France, in 1999.

He is currently a Principal Scientist in the PowerElectronics Group at ABB Switzerland, Ltd. Zurich,Switzerland. His main research interests include non-

linear control design, passivity based control, control of switching power con-verters, active filters, inverters, electrical drives, and renewable energy systems.

Leonardo Augusto Serpa (M’04) received the B.S.and M.S. degrees in electrical engineering from theFederal University of Santa Catarina, Florianopolis,Brazil, in 2002 and 2004, respectively, and the Ph.D.degree in power electronics from the Swiss FederalInstitute of Technology (ETH), Zurich, Switzerland,in 2007, and the MBA degree from Cass BusinessSchool, London, U.K., in 2011.

Between July and December 2001, he was anIntern at the Center for Power Electronics Systems(CPES) in Virginia. From 2007 to 2010, he was a

Principal Scientist at ABB Corporate Research, Baden, Switzerland, where heled research projects on power electronics for renewable energy sources. He iscurrently an Assistant Vice President at the Corporate Strategy Group of ABBManagement, Zurich, Switzerland.

Antonio Coccia (M’06) received the Master andPh.D. degrees from the University of Naples, Fed-erico II, Italy, in 2001 and 2004. His Ph.D. studieswere jointly funded by AnsaldoBreda TransportationSystems, Naples, where he worked on power elec-tronic converters and drives for traction applications.

In 2005, he joined ABB Corporate Research,Baden, Switzerland, as a Scientist in the Power Elec-tronics Systems Group. He was promoted to firstPrincipal Scientist and then to Group Manager of thePower Electronics Systems and Applications Group

in 2007. Since August 2010, he is Department Manager in Vehicle and SystemsEngineering together with Bombardier Transportation/Locomotives Division.

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages false /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/Description >>> setdistillerparams> setpagedevice

ieee transactions on power electronics, vol. 27, no. …ho et al.: practical design and...

Documents